Hello Kundan,
Thanks for your detailed comments. We will come back to you with our opinions by today evening.
Best Regards Hemant Agrawal Vipin Palawat
----- Original Message ----- From: Kundan Singh kns10@cs.columbia.edu To: Hemant Agrawal hemantag@graffiti.net Cc: schulzrinne@cs.columbia.edu; jdrosen@dynamicsoft.com; korpim@sbs.de; karl.klaghofer@vs.siemens.de; rrroy@ATT.COM; alan.johnston@wcom.com; steven.r.donovan@wcom.com; Kevin.Summers@wcom.com; dean.willis@wcom.com; henry.sinnreich@wcom.com; VIPIN PALAWAT vipin.palawat@wipro.com; agboh@helios.iihe.ac.be; Robert.Sparks@wcom.com; Chris.Cunningham@wcom.com; orit@radvision.com; taylor@NORTELNETWORKS.COM; joon_maeng@vtel.com; drwalker@ss8networks.com; paul.jones@TIES.ITU.INT; adilber@ATT.COM; stephen.terrill@ericsson.com Sent: Wednesday, March 08, 2000 11:34 PM Subject: Re: SIP-H.323 Interworking Call Scenarios
Hi Hemant and Vipin,
Its a very good collection of call flows' translation.
Few comments:
- Might want to include H.323---SIP---H.323 case where the
call goes through two SGWs. Since, combining the two call flows (H.323-SIP, sec:1.1) and (SIP-H.323, sec:2.1) may not work. The SIP ACK in sec:1.1 has a different session description, which will trigger some "Mode Request" and/or "Open/Close LogicalChannel" procedures in sec:2.1.
In sec:1.1,1.2,1.3; it might be better to send ACK immediately after receiving 200 OK, and send re-INVITE once H.245 procedures are complete. This way, we can avoid the retransmissions of 200 OK by SIP EP if H.245 procedures take more time to complete. Comments ??
In Fast Connect procedures (sec 3.1), I assume that if the normal H.245 procedure resumes after initial call setup (with faststart), and there is a change in H.323 side session description, then appropriate re-INVITE message is sent to SIP EP.
Sec:4 Call transfer may be renamed to (Blind) Call Tranfer.
Sec:4.1. RTP should be between SIP EP(B) and SIP EP(C)
Sec: 5.1, you may send "181 Call is being forwarded" response to SIP (A). Can SIP (A) reject the call forwarding here ??? Who pays for the call if the forwarded call costs more (longdistance/international) ? Any pointer to this topic is appreciated.
The call forwarding may be initiated by the proxy (after receipt of 486 Busy Here) or by the SIP endsystem itself (by responding with 302 Moved Temporarily) if it is configured to do so. However, that will not affect the translation much.
Sec:5.3, typo; SIP User (A) should be (B) and (B) should be (C).
Considering the wide variety of scenarios between H.323 and SIP, would it be reasonable to start with a simple call translation specification, provide a state machine/pseudo code for message handling by the SGW (signaling gateway) and then extend the state machine/pseudo code to handle all the non-trivial call scenarios.
Once the basic translation specification is ready, profiles for extensions (e.g., blind transfer using GK, call forwarding using proxy, translating forking proxy behaviour to H.323, etc.) may be specified.
Thus, the initial specification may cover following areas:
- Simple SIP--SGW--H.323 translation, with SGW independent of any GK or proxy.
- Translating (blind) call transfer by endsystems.
- Translating call forwarding by endsystems.
- How to handle, SIP re-INVITEs (change in session description/or media transport address) and change in H.323 logical channel/mode request.
- How to handle H.323 fast start.
Later on other things may be added:
- Call hold translation
- Call transfer/forwarding by GK and/or proxy.
- SIP--H.323--SIP and H.323---SIP---H.323 scenarios.
- SGW coexiting with GK or proxy.
...
Motivation for such a separation is to allow the implementors more freedom in deciding what configuration to choose, while still maintaining the basic translation framework. For example, call forwarding to voicemail may be handled by
- the endsystem by sending 302 moved, after say 4 rings.
- the proxy server on receipt of 486 busy here
- the proxy server, forking one branch of the request to the voicemail
system which accepts the call after, say, 15 seconds.
It turns out that once we have the initial specification, (as mentioned in above list) the advanced scenarios are just a "special cases"/"combination of cases" of the initial scenarios.
Regards
Kundan
Kundan Singh http://www.cs.columbia.edu/~kns10
On Wed, 8 Mar 2000, Hemant Agrawal wrote:
Hi All, There was a continuous discussion on the different configuration for the SIP - H.323 interworking.
In our opinion, it would be better if some basic features (like Call Transfer, Call Hold and Call forwarding) is also added as part of the discussion.
We are enclosing some call flow examples of SIP-H.323 interworking.
Please
give your comments on these call flows. Our stress is on the call
features
and others are just informational. We have specified few issues also
inorder
to make these protocols interwork.
Kindly let us know if you agree to include the feature discussion also
as
part of the SIP - H.323 interworking. We are ready to put our efforts
for
the feature interworking.
Best Regards
Hemant Agarwal Vipin Palawat